On the dialog between experimentalist and modeller in catchment hydrology: Use of soft data for multi-criteria model calibration

Table of Contents

On the dialog between experimentalist and modeller in catchment hydrology: Use of soft data for multi-catchment model Calibration

Experimentalists versus modelers

What is "soft data"?

Slide 4

3-box model

Types of soft data

Dialog between experimentalist and modeller

Evaluation rules

Model performance

Model efficiency: 0.93

Model efficiency: 0.92

Model efficiency: 0.93

Best overall performance

Reduction of parameter uncertainty

Concluding remarks

Final remark

Objectives of this talk

Evaluation rules

Slide 19

Home Page: http://www.cof.orst.edu/cof/fe/watershd/

Abstract
The use of field data for model calibration is often limited beyond the use of streamflow information despite the general acceptance that internal state informal are necessary for ensuring internal model consistency. Hydrologists often have a highly detailed yet highly qualitative understanding of dominant runoff processes-thus we usually know much more about a catchment than we use for calibration of a model. We present a new method where weak information (i.e., qualitative knowledge from the experimentalist, which cannot be used as exact numbers) is made useful through fuzzy measures of model-simulation and parameter-value acceptability. A three-box model was developed for the Maimai catchment in New Zealand, a particularly well-studied process-hydrological research catchment. The boxes represent riparian, hollow and hillslope zones. These zones differ significantly in their groundwater dynamics, physical soil characteristics, stable isotope composition and end member chemistry. The model was calibrated against hard data (runoff and groundwater-levels) as well as a number of criteria derived from the weak information. Parameter sets were evaluated via: (1) the traditional comparison of simulations with observations for hard data such as time series of runoff and groundwater levels, (2) the acceptability of the parameter values with respect to field knowledge (e.g., size of the riparian zone or ratio of the hydraulic conductivity in the riparian zone compared to that in the hollow zone) and (3) the acceptability of the simulations with regard to field observations (e.g., size of events with groundwater occurrence in hollow zone or peak groundwater levels in the different zones). Using a comparatively large number of criteria for model evaluation helped to reduce parameter uncertainty and to ensure a better process representation. The proposed method of using weak information for model calibration and validation is also a way to encourage dialogue between the modeler and the experimentalist. It is also useful for comparing the value of different field measurements in support of modeling.